Segmentation of lines based on point densities--an optimisation of wildlife warning sign placement in southern Finland.

Current and previous placement of wildlife warning signs by the Finnish Road Administration is based on suggestions from hunting associations and individuals. The problem within this practice is that the placement of warning signs is not a transparent process. Within this paper, we evaluate the current placement of wildlife warning signs for the Uusimaa region in southern Finland and develop a method to optimise warning sign placement using kernel density estimations, which are based on existing wildlife-vehicle accident records. The contour lines of these densities can indicate the road sections to be marked with warning signs. To apply a well-documented computational method, based on these moose and white-tailed deer accident locations, assists the Finnish Road Administration in their task to place, replace and assess wildlife-warning signs for specific road sections.